python/etc/fqdeumi.py

   1 #!/usr/bin/env python3
   2
   3 import sys
   4 import os
   5
   6 UMI_LENGTH = 9
   7
   8 def process(lines=None):
   9     ks = ['name', 'sequence', 'optional', 'quality']
  10     return {k: v for k, v in zip(ks, lines)}
  11
  12 try:
  13     fn = sys.argv[1]
  14 except IndexError as ie:
  15     raise SystemError("Error: Specify file name\n")
  16
  17 if not os.path.exists(fn):
  18     raise SystemError("Error: File does not exist\n")
  19
  20 n = 4
  21 with open(fn, 'r') as fh:
  22     lines = []
  23     for line in fh:
  24         lines.append(line.rstrip())
  25         if len(lines) == n:
  26             record = process(lines)
  27             #record['umi'] = record['sequence'][:9]
  28             record['name'] = ' '.join((record['name'][1:],record['sequence'][:UMI_LENGTH]))
  29             record['sequence'] = record['sequence'][UMI_LENGTH:]
  30             #sys.stdout.write("Record: %s\n" % (str(record)))
  31             print(''.join(('>',record['name'])))
  32             print(record['sequence'])
  33             lines = []